Local-first personal multi-LLM memory toolchain and MCP server.
This repository also configures its own agents. See what ai-dememory tells them →
Skill Claude CodeCodex
Inspect or run bounded ai-dememory daily and weekly upkeep, provider imports, generated indexes, recall quality, review queues, and scheduler health. Use when the user asks for maintenance, autonomous upkeep, resource consumption, cleanup, or schedule status.
Skill Claude CodeCodex
Recall focused, reviewed context from ai-dememory with bounded MCP calls. Use before non-trivial work when prior decisions, project context, preferences, values, or recommendations may materially change the result, or when the user asks to remember or search memory.
Skill Claude CodeCodex
Review ai-dememory inbox proposals, provider imports, explicit captures, hook captures, recall misses, false positives, and conflict candidates. Use when the user wants to promote, reject, clean, consolidate, dismiss, or write conflict merge proposals.
Skill Claude CodeCodex
Set up ai-dememory for a local vault, MCP, bounded hooks, provider imports, or reviewed maintenance schedules. Use when the user asks to install, configure, run the initial wizard, choose resource intensity, or enable autonomous local upkeep.
Skill Claude CodeCodex
Maintain generated working state and handoffs from ai-dememory through MCP or CLI tools. Use when the user asks to remember current task state, prepare a handoff, resume recent work, summarize a session, or update local working memory without promoting durable facts.
Skill Codex
Use when Codex needs to operate the installed ai-dememory tool against an explicitly bound private vault: recall reviewed Markdown memory, validate or index that vault, run safe secret checks, export bounded context, write proposal captures, review inbox entries, or operate the local MCP server. Also use separately…
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: